NEURAL NETWORK BASED REINFORCEMENT LEARNING ACCELERATION ON FPGA PLATFORMS

Add to your list(s) Download to your calendar using vCal

• Jiang Su
• Friday 22 July 2016, 14:00-15:00
• Mahanakorn Room, EEE.

If you have a question about this talk, please contact Joshua M Levine.

This is the same conference practice talk with the one that was given on Monday. But the slides have been massively improved and the presentation will be restricted in 20 minutes. Please feel free to attend if you are interested in the topic but somehow confused by the talk on Monday. The abstract of the work is shown below:

Deep Q-learning (DQN) is a recently proposed reinforcement learning algorithm where a neural network is applied as a non-linear approximator to its value function. The exploitation-exploration mechanism allows the training and prediction of the NN to execute simultaneously in an agent during its interaction with the environment. Agents often act independently on battery power, so the training and prediction must occur within the agent and on a limited power budget. In this work, We propose an FPGA acceleration system design for Neural Network Q-learning (NNQL). Our proposed system has high flexibility due to the support to run-time network parameterization, which allows neuro-evolution algorithms to dynamically restructure the network to achieve better learning results. Additionally, the power consumption of our proposed system is adaptive to the network size because of a new processing element design. Based on our test cases on networks with hidden layer size ranging from 32 to 16384, our proposed system achieves 7x to 346x speedup compared to GPU implementation and 22x to 77x speedup to hand-coded CPU counterpart.

This talk is part of the Jiang Su's list series.

This talk is included in these lists:

• Mahanakorn Room, EEE

Note that ex-directory lists are not shown.